基于K近邻改进算法的城市配送量预测研究  

Urban Logistics Distribution Volume Forecast Based on Improved K-nearest Neighbor Algorithm

作　　者：肖赟 刘洋 裴爱晖[3] 梁子君 XIAO Yun;LIU Yang;PEI Ai-hui;LIANG Zi-jun(School of Urban Construction and Transportation,Hefei University,Hefei 230601,China;Anhui Province Transportation Big Data Analysis and Application Engineering Laboratory,Hefei 230601,China;Research Institute of Highway Ministry of Transport,Beijing 100088,China)

机构地区：[1]合肥学院城市建设与交通学院,合肥210601 [2]安徽省智慧交通大数据分析与应用工程实验室,合肥210601 [3]交通运输部公路科学研究院,北京100088

出　　处：《淮阴工学院学报》2022年第3期1-7,30,共8页Journal of Huaiyin Institute of Technology

基　　金：安徽省自然科学基金(2208085ME147);中央级公益性科研院所基本科研业务费专项资金项目(2020-9071)。

摘　　要：准确预测城市配送订单量对于科学调配城市配送资源和优化平台运营管理具有重要意义。以合肥市城市配送平台2018-2020年订单数据为例,以7天为时间周期组成数据集,对订单历史数据库进行标准化清洗,以平均绝对误差最小为优化目标,优化了K近邻模型的参数K和状态向量T的取值。研究表明:当K=2,T根据不同数据类型取值1~4时,预测误差最小。对比分析历史平均模型与移动平均算法,三种预测方法的平均绝对百分比误差分别为8.34%、23.65%、11.16%,K近邻算法的预测精度高于其它两种方法,预测精度接近91.7%,能够较好地预测城市配送订单需求量,具有较良好的应用推广价值。Accurately predicting the volume of urban distribution orders is of great significance for sci⁃entifically allocating urban distribution resources and optimizing platform operation management.The pa⁃per took the order data of the Hefei urban distribution platform from 2018 to 2020 as an example,a data set was formed every 7 days,and its order history database was standardized and cleaned,and we also optimized the parameters of the K-nearest neighbor model with the minimum average absolute er⁃ror as the optimization goal.The parameter K and state vector T of the K-nearest neighbor model were optimized.The conclusion showed that when K=2 and T was between 1~4 according to different data types,the prediction error was the best.The historical average model and the moving average algo⁃rithm were compared and analyzed.The average absolute percentage errors are 8.34%,23.65%,and 11.16%,respectively.The prediction accuracy of the K-nearest neighbor algorithm is higher than the other two methods,whose accuracy is close to 91.7%.The model can better predict the demand of ur⁃ban distribution orders,and has good application and promotion value.

关 键 词：城市交通 订单预测 K近邻 城市配送 

分 类 号：U491[交通运输工程—交通运输规划与管理]